global {
    usage-count yes;
}
common {
    syncer {
        rate 1M;
    }
}

resource DRBDCluster {
    protocol C;
    net {
        cram-hmac-alg sha1;
        shared-secret "sdfdskljdfklfjsdkljfsdkljflk";
    }
    on srv1.intranet.example.org {
        device    /dev/drbd0;
        disk      /dev/sdb1;
        address   1.1.1.1:7789;
        meta-disk internal;
    }
    on srv2.intranet.example.org {
        device    /dev/drbd0;
        disk      /dev/sdb1;
        address   1.1.1.2:7789;
        meta-disk internal;
    }
}

14 thoughts on “DRBD on Red Hat Enterprise Linux 5”

1. Pingback: Tweets that mention Sébastien Wains » DRBD on Red Hat Enterprise Linux 5 -- Topsy.com

2. Hi Sébastien, thank you.

   The default sync rate in debian DRBD packages and fedora is 100M. Should I set my own rate?

   Have you own calculation rules?

   Have you a own rule for “on-io-error”, detach? or something special?

   Here is a calculation guide for sync rates: http://www.drbd.org/users-guide/s-configure-syncer-rate.html

   • There’s no exact science with this setting. It really all depends on the application, how much data has to be replicated, if the nodes are on the same network/premises or not.

     The default value used to slow down a Zarafa Cluster. HP G4 nodes were two separate physical locations distant from around 5 kms, with a Gigabit link.

     I’d say that this setting has to be worked out during the test phase. Decrease the value until the application is responsive in high load tests.

3. Thank you very much for the explanation:)

   • My pleasure

     Cool blog, glad you liked my OpenVPN posts
     OpenVPN rocks eh ?

4. Hey:) your posts are great! not only about OpenVpn:)

   And yeah it rocks. no downtime – no problems with dynamical ips, I love it:)

5. Sébastien what was your solution to access zarafa web-acces? I mean you have 2 separated servers with different locations. Have you a special solution for the virtual-IP?
   _ _ _ _ _ _ _ _ _
   srv1.intranet.example.org has a network A – 7.7.7.11
   srv2.intranet.example.org has a network B – 8.8.8.10
   What I mean you can’ t have a virtual IP with different networks?
   - – - – - – - – - – - – -

   I think you have a virtual IP-Address to access the running server, but what was the solution to bring it to work with different networks and a virtual IP?

   By the way what is your cluster stack?

   Thank you!

   • For the gateway side (MTA, webmail, IMAPS, POP3S) we are using a Cisco ACE load balancer doing SSL offload. With this solution it is possible to load balance between 2 separate networks, as long as the load balancer can reach those two real servers.

     For the Zarafa side, we are using Cisco ACE along RHCS in failover mode. Cisco ACE VIP is publicly available, RHCS VIP is internally available only. Cisco ACE is somewhat doing DNAT here.

     I guess srv1 and srv2 are internal networks and you want to make the webmail publicly accessible, right ? But do you want to load balance or failover ?

6. I’m working currently on a failover solution – for 2 servers with different locations. That means I have a Problem with my virtual IP. It will be a web-server with cluster filesystem (can’t decide which one:) – maybe you can recommend something? OCFS2?) to run mysql and apache on each node under DRBD.

   Example:
   http://www.example.com –> vhosts.example.com
   vhosts.example.com –> srv1.example.com and srv2.example.com

   If one server goes down, everything should be taken by another node. My provider (hetzner.de) sells Failover-IP which can be variable assigned to a server. This is the only one solution I see.

   Thank you for a quick reply.

   • For the MySQL part, we have great success running it over DRBD. I would stick to this solution personally.
     I have made some tests with a failover Apache cluster over RHCS and DRBD. It works fine and easy to set up.
     I tend to avoid anything involving Oracle, never played with OCFS2.

     Take a look at IPVS and Keepalived. I haven’t played with those in years, but I believe it should fit. Let me know if it does, or if I’m completely mistaken

7. hi wains,

   i receive this error. what does it mean.

   Found LVM2 physical volume signature
   Device size would be truncated, which
   would corrupt data and result in
   ‘access beyond end of device’ errors.
   If you want me to do this, you need to zero out the first part
   of the device (destroy the content).
   You should be very sure that you mean it.
   Operation refused.

   thanks
   nazrin

8. hi wains,

   i have figure out the problem for above comment. now i got issue when try to start for drbd services…

   [root@primary testsvr]# /etc/rc.d/init.d/drbd start
   Starting DRBD resources: [ d(repdata) /dev/drbd0: Failure: (114) Lower device is already claimed. This usually means it is mounted.

   [repdata] cmd /sbin/drbdsetup /dev/drbd0 disk /dev/sda /dev/sda internal –set-defaults –create-device –on-io-error=detach failed – continuing!

   ].

9. you’re obviously not following the steps. You can easily get errors if using a command before another. Start over

10. retry again today…

Comments are closed.